Establishment Of Recurrence Prediction Model After Radical Gastrectomy For Gastric Cancer

Objective Through collating clinicopathological data of patients after radical gastrectomy for gastric cancer,the risk factors were explored,the prediction model for recurrence was established,and the prognostic analysis was carried out,aiming to provide basis for individualized treatment and follow-up of patients after radical gastrectomy for gastric cancer.Methods A retrospective study was conducted to collect clinicopathological data of patients who underwent radical gastrectomy for gastric cancer from March 1,2017 to March 31,2020 in the Department of Oncology,First Hospital of Lanzhou University.The recurrence of patients was determined by medical records and telephone followup.According to the recurrence,patients were divided into recurrence group and nonrecurrence group.Univariate and multivariate analyses of clinicopathological features were performed by binary logistic regression to find the independent risk factors for gastric cancer recurrence.A recurrence prediction model was established and corrected based on the independent risk factors.Kaplan-meier survival curve was used to analyze the prognostic analysis of statistically significant factors in univariate analysis.Results The study included 300 gastric cancer patients with a follow-up deadline of August 23,2021,with a median follow-up time of 26.07 months,and the median recurrence free survival（RFS）was not reached.There were 105 recurrent patients,with a recurrence rate of 35%（105/300）.Univariate analysis showed that CEA at first diagnosis,CA199 at first diagnosis,CA724 at first diagnosis,tumor size,pathological type,degree of differentiation,whether vascular involvement,lymphatic involvement,T stage,N stage and p TNM stage were correlated with tumor recurrence after radical gastrectomy.Multivariate analysis showed that CA199,tumor size and p TNM stage were independent risk factors for gastric cancer recurrence.Three factors including CA199,tumor size and p TNM stage were used to establish the recurrence prediction model,and the Receiver Operator Characteristic（ROC）curve was drawn to evaluate the recurrence prediction model.The area under the curve was calculated to be 0.828,suggesting that the accuracy of the model was high.According to the relationship between the predicted probability and the actual probability,the correction curve obtained by the prediction model coincides with the ideal curve,which proves that the model has good prediction efficiency.Survival analysis showed that CEA at first diagnosis,CA199 at first diagnosis,CA724 at first diagnosis,tumor size,whether vascular involvement,lymphatic involvement,T stage,N stage and p TNM stage were important influencing factors for prognosis of gastric cancer patients.Conclusion The recurrence rate after radical gastrectomy is high.There were three independent factors affecting recurrence of gastric cancer,namely CA199 at first diagnosis,tumor size and TNM stage.The established recurrence prediction model can better predict the recurrence risk.The prognostic factors of gastric cancer include CEA at first diagnosis,CA199 at first diagnosis,CA724 at first diagnosis,tumor size,pathological type,whether vascular involvement,whether lymphatic involvement,T stage,N stage and p TNM stage.